Accelerate Your Digitization with AI
Interactively segment your rasters in QGIS. Spend less time tracing, more time analyzing.

AI Segmentation
Transform how you digitize.
Point at objects. Get precise vectors. Ship faster.
Manual digitization
We all know the pain.
It's long. Vectorizing buildings, parcels, or vegetation on orthophotos can take hours, even days.
It's repetitive. Tracing the same type of object over and over — exactly the kind of task we'd love to automate.
It's frustrating. Between endless zooming, missed clicks, and complex contours, manual digitization demands patience.
What if we could do it differently?
Interactive Segmentation
AI-assisted extraction.
Point at objects. Get precise contours. Instantly.
AI Segmentation uses Segment Anything Model (SAM), a state-of-the-art deep learning model. Instead of tracing manually, you simply click — left to include, right to exclude — and the AI generates precise vector polygons.
The result? What used to take 2-3 minutes per object manually now takes a few seconds.
Where it shines
🏘️Urban Planning & Cadastre
Update cadastral databases after new constructions. Segment buildings on recent orthophotos, export to vector, import into your GIS. Half a day → 1-2 hours.
🌾Precision Agriculture
Analyze agricultural plots from drone imagery. Segment homogeneous zones — crops, stress areas, weeds — for prescription maps. No tedious manual work.
Built for QGIS
Fits your workflow.
Automatic Installation
Dependencies install on first launch.
Intuitive Interface
Left to include, right to exclude, Ctrl+Z to undo.
Multi-platform
Windows, macOS (Apple Silicon), Linux.
Stable Architecture
PyTorch runs in a separate process.
Direct Export
Save as GeoPackage with attributes.
Universal Compatibility
Satellite, drone, aerial, scanned maps.
What you need
| QGIS | 3.0+ |
| RAM | 8 GB (16 GB rec.) |
| Disk | ~2 GB |
| OS | Win · macOS · Linux |
| Internet | Initial install only |
Get Started
Need a Custom Solution?
From custom segmentation models to full pipeline integration — we build what doesn't exist yet.